Instrument Calibration and Radiance Validation of GOES-R ABI

1. Instrument Calibration and Radiance Validation of GOES-R ABI Fred Wu, October 18, 2018

2. Acknowledgements • Current CWG members: F. Yu, V. Kondratovich, H. Yoo, H. Qian, Z. Wang, and S. Guo. • Former CWG members: C. Cao, F. Padula, A. Pearlman, D. Pogorzara, R. Datla, X. Shao, M. Grotenhuis, M. Chu, B. Efremova, R. Iacovazzi, L. Zhu, J. Choi, and R. Mirza-Saliq. • External members: T. Schmit, M. Gunshor, D. Lindsey, M. Coakley, K. Mitchell. • Colleagues at Harris: J. Van Naarden, D. Gall, L. Roop, W. Harting, M. Wilson. • Colleagues with GOES-R Program: C. Rollins, A. Reth, J. Fulbright, E. Kline, W. Labair, J. Kronewetter, M. Seybold, E. Grigsby, S. Goodman, C. Keeler, A. Krimchansky. • Colleagues in STAR: J. Daniels, AWG leads, L. Brown, S. Kalluri, Front office. • And many more. GOES-R ABI Cal Val

3. Calibration, Verification, and Validation • Calibration • Instrument Calibration: Characterization of instrument responses to known signals. • Measurement Calibration: Conversion of instrument response to physical quantity. • Product Calibration: Homogenization of multiple products, e.g., inter-calibration of radiance; empirical correction of higher level products. • Verification: System did what it is supposed to. • Validation: System achieved what it is intended to. • “Instrument Calibration and Radiance Validation …” GOES-R ABI Cal Val

4. History of GOES ImagingSpace weather, solar observation, sounding not covered • 1st Generation: • GOES-1/2/3. October 16, 1975. • Visible Infrared Spin Scan Radiometer (VISSR) • 1.5 (or 2nd) Generation • GOES-4/5/6/G/7 • VISSR Atmospheric Sounder (VAS) • 2nd (or 3rd) Generation: • GOES-8/9/10/11/12, April 1994 • Imager (and Sounder) • 3-axis stabilized platform • 2.5 (or 4th) Generation: • GOES-13/14/15/Q • Improved batteryand navigation. • 3rd (or 5th) Generation: • GOES-16/17/T/U, 19 Nov 2016 • Advanced Baseline Imager (ABI) • ~500 times more detectors. GOES-R ABI Cal Val

5. Past Generation Transition 3-axis stabilization – AOL Variation M. Weinreb GOES-R ABI Cal Val

6. Past Generation Transition 3-axis stabilization – Temperature Variation One Week One Month ~60ºK some other days ~40ºK “now” Complicated seasonal variations of diurnal heating Long term trend D. Han One Year One Satellite Goes-12 Scan Mirror Temperature GOES-R ABI Cal Val

7. Past Generation Transition 3-axis stabilization – Temperature Variation TBB TBB OK if =1 orTenv=TBB Less true for GOES Tenv M. Weinreb (Modified) GOES-R ABI Cal Val

8. Past Generation Transition I. Chang GOES-R ABI Cal Val

9. Expected GOES-R Benefits … • Compared to Imager, ABI must provide … • 3 times more channels (5 –> 16) • 4 times finer spatially (4/1 km –> 2/0.5 km) • 5 times faster temporally (26 min –> 5 min) • … and the data must be better. • Optimized (narrower) SRF • Lower noise • Calibrated solar bands • More detectors – 16 –> 7,856 (24 –> 77,400). Harris Corp & T. Schmit GOES-R ABI Cal Val

10. … Come With New Challenges … • Large arrays of detectors – Spectral Uniformity. • Multiple focal planes – Geometric calibration based on Kalman filter • From “aim then collect” to “collect then tell” • Radiometric and geometric calibrations become more closely intertwined. • Cooling – Loop Heat Pipe • Onboard calibration for the solar bands. • New telescope design – Spatial Uniformity. GOES-R ABI Cal Val

11. … Also New Tools … • Global Space-based Inter-Calibration System (GSICS) – rapid validation. • Lunar Calibration – thanks to GSICS partners. • Rayleigh Scattering – new channel. • North-South Scan – new capability. • CRTM GOES-R ABI Cal Val

12. … Drive PLPT Plan … Critical tests are in bold. Experimental tests are in italic. GOES-R ABI Cal Val

13. … Captured in RIMP • Planning • Readiness, Implementation, and Management Plan • Execution • Post-Launch Tests (PLT) – Beta • Verification of instrument for required functionalities • Planned adjustments • Flight-Ground Integration • Post-Launch Product Tests (PLPT) – Provisional • Validation of products for intended applications • Program-User Integration • Extended Validation (XV) – Full Validation • Comprehensiveevaluation of operation in intended environment • Phase into normal operation GOES-R ABI Cal Val

14. VNIR Noise • Meet the requirement with large margin. • Mean has been stable. • 064 Band is challenging. GOES-R ABI Cal Val

15. VNIR Noise Stability GOES-R ABI Cal Val

16. IR Noise GOES-R ABI Cal Val

17. IR Noise Stability GOES-R ABI Cal Val

18. VNIR Accuracy GOES-R ABI Cal Val

19. VNIR Accuracy Stability Outliers are due to GS, e.g. the 1.61μm before Feb. 2018. GOES-R ABI Cal Val

20. IR Accuracy GOES-R ABI Cal Val

21. IR Accuracy Stability

22. Navigation Error (I) Many did not meet the requirement at Provisional • By a lot for some bands. All met the requirement at Full Validation • With large margin. GOES-R ABI Cal Val

23. Nav Improvement – Residual Errors were reduced by an order of magnitude in a year, after halved before then. GOES-R ABI Cal Val

24. Nav Improvement – STD STD were reduced to 1/3. Total Error = Residual + 3*STD GOES-R ABI Cal Val

25. Navigation Error (CCR) … and better than required, but some not as good as expected. GOES-R ABI Cal Val

26. Navigation Error (FFR) • For the window channels only. • Sounding channels should show FFR similar to the window channels on the same FPA. • Better than requirements and baseline since provisional. • Still, notable improvement in last year. GOES-R ABI Cal Val

27. PICA – Periodic Infrared Calibration Anomaly • PICA is the anomaly with 15-minute of periodicity observed at GOES-16 ABI Mode-3 MESO and CONUS images. • Investigation of the space MESO images from the lunar chasing events revealed root cause was due to the residuals of incident-angular scan-mirror emissivity corrections • Update the scan-mirror emissivity look-up-tables (LUTs) on 10/19/2017 greatly reduced the PICA impact, improved the IR spatial uniformity, reduced the diurnal calibration variations, and ensured more consistent GEO-LEO inter-calibration results to CrIS and IASI. PICA impact. Courtesy of C. Gravelle via D. Lindsey Reduced PICA impact after GS update More consistent GEO-LEO Tb biases Improved spatial uniformity Reduced diurnal variation GOES-R ABI Cal Val

28. B01-03 Striping • Striping was observed at GOES-16 ABI VNIR bands, especially at B01/02/03 since PLT period • CWG reported discrepancy of Q-scaling scheme between the operational calibration algorithm and the Equation described in radiometric calibration ATBD in Sept. 2017 • Vendor believed silicon detector response may also get changed in-orbit (e.g. ~4% jump of B02 in mid-March 2017) • New Q values derived iteratively with in-orbit Earth data for B01-B03 using Q-scaling algorithm • The new Q values for B01/02/03 and “Q-scaling” scheme for all the six VNIR bands were implemented completely on 02/07/2018. • VNIR B01/02/03 striping significantly reduced • B04 experienced ~3% reduce in the radiance while change in the other channels was less than 1.5% GOES-16 ABI B02 images obtained at 11:00Z (left) and 11:25Z (right) on 7 February 2018. (image provided by T. Schmit) showing the reduction in image striping after the complete implementations of the striping reduction parameters and algorithms at 11:16Z Mean radiance difference estimated for the Earth scenes before and after the operational algorithm update GOES-R ABI Cal Val

29. Solar Calibration Timing • SD reflectance varies bi-directionally. • Pre-launch, it was known for one elevation angle (0°) and variety of beta angle (-23.5°). • Solar calibration should occur when elevation angle is zero. • In 2017, solar cal often occurred at the fixed clock time with varying elevation angle. • Negative impact on solar calibration • Fixed since Dec 2017. GOES-R ABI Cal Val

30. Comments • Anomalies resolved before Provisional were often due to operational errors • Reversed order of LUT; • Incorrect Integration Time Multiplier; • Outdated files after CBU switch; etc. • Anomalies resolved after Provisional require comprehensive understanding of ABI instrument calibration, ingenuous analysis, and deep dive into the data. • Will continue into future. GOES-R ABI Cal Val

31. Latched Detector Response • ABI IR detectors may experience sudden change in response – detector level shift • The level shift, if greater than lunar rejection threshold, can set the space look of this detector “latched”. Striping thus occurs • Erroneous calibration coefficient • Striping is removed or reduced: • If response (space look and ICT counts) is not saturated, striping is removed by resetting the space look statistics object. • If response is saturated, space look reset can trigger “band-aid” algorithm to reduce apparent striping till the BDS is updated. • Ongoing effort for the current operational algorithm to protect from the saturated ICT Actual detector responses Detector level shift “Latched” space look used for operational calibration GOES-R ABI Cal Val

32. 11.2μm Striping • Horizontal striping has been detected for the 11.2μmband and, less frequently, for some other IR bands. • For the 11.2μmband, the most obvious striping is linked to 3 detectors with non-zero Q. The same correlation is discovered for bands 15 & 16. • The striping patterns also suggest that the Q LUT of these detectors are probably problematic. • There is an on-going effort by CWG to determine the root cause. repeat each swath D307 D250 D176 GOES-R ABI Cal Val

33. ICT PRT Coefficients • In April 2018, a possible discrepancy between the ICT PRT coefficients saved in Flight CDRL-42 and GS LUT was discovered and later confirmed. • The implementation of the correct CDRL 42 coefficients shall shrink the spread of the 12 ICT PRT temperatures from ~ 6 K to < 0.1 K. The reported average ICT temperature for gain calculation shall increase by 0.2 K. • The update of GS LUT is expected to remove at least the majority of ~ -0.2 K bias observed for all IR bands.  PRT Coeff. from GS LUT PRT Coeff. from CDRL 42 GOES-R ABI Cal Val

34. Cold Pixels Around Fire (CPAF) • Compare 390um channel with a longwave IR channel such as 112um for a fire scene, the fire can be much hotter in 390 band than 112 band (412K or saturated vs. 302K). • Nearby pixels can be much colder in 390 band than 112 band (0K or saturated vs. 287K). • Both phenomena are well understood. • Flag has been implemented in operation, and works well, to avoid confusion. • User requested enhancement, which seems within reach. • Several options have been considered. One was chosen. Work is underway. GOES-R ABI Cal Val

35. 064 Band Bias • 064 Channel is more than 5% brighter than pre-launch, VIIRS, and AVIRIS-NG (Field Campaign). • It is likely caused by uncertainty in the K factor, the detector- and angle-dependent reflectance of solar diffuser that was determined pre-launch. • There is no consensus whether K is wrong and, if it is, what is the right value for K. • There is consensus that the calibrated radiance is wrong and users demand quick change. • It is recommended to adjust K to bring 064 radiance to be in family. GOES-R ABI Cal Val

36. Strong Refracted Solar Light Leaking onto ABI Detector in March, 2018 • At night side, strong solar light leaked onto ABI detectors when Sun is behind Earth due to refraction of solar light propagating through the atmosphere • Occurred frequently when the sun is near the limb margin in March, 2018 • Caused the ABI VNIR detectors to saturate. Harmful for ABI detectors. • ABI solar avoidance truncation was confirmed to perform correctly with the limb margin set at 8.1 degree • It was concluded that the swath truncation is not adequate with the current limb margin at 8.1 degree • Proposed adjustment of limb margin radius to 7.9 degree or lower to avoid the strong leaking light. The MODTRAN 2/3 Report and LOWTRAN 7 MODEL GOES-R ABI Cal Val

37. Proposed Solar Avoidance Parameter to Avoid Strong Refracted Solar Light Leaking onto ABI Detector Mar. 3, 2018, ABI CH03 Mar. 2, 2018, ABI CH03 Mar. 19, 2018, ABI CH07 Current Swath Truncation with Limb Margin at 8.1 Degree Missed Missed Missed Proposed Swath Truncation with Limb Margin at 7.9 Degree • The proposed change of the limb margin radius to 7.9 degree for solar avoidance truncation can effectively avoid the strong leaking solar light onto the ABI detectors GOES-R ABI Cal Val

38. Summary • Definition of cal/val. • Cal/val lessons from past generation transition. • Planning of ABI cal/val • Execution of ABI cal/val • Key results of ABI cal/val GOES-R ABI Cal Val